Sie sind auf Seite 1von 10

HPLC separation of fermentation products glycerol and 1-3 propanediol

Transport Lab Spring 2006 Team 10


Vincent Chou Corey Mays Robert Wong Carmen Jones

Abstract

Fermentation of corn starch is a method of producing 1-3 propanediol (PDO) from glycerol. To monitor the 1-3 propanediol levels the fermentation products are put through high pressure liquid chromatography (HPLC). However, these two molecules

have very similar in structure and physical properties. So to generate better separation of the two major products of the fermentation process the flow rate and temperature were changed and implemented on known mixtures of these two major products. These experiments agreed with the findings about how these factors would effect the separation and helped the group get more acquainted with the HPLC machine.

Table of Contents

Introduction1 Background and Theory1-4 Equipment4 Procedure4-5


2

Results and Discussion5-7 Future Recommendations7-8 Conclusion8 References 9

Introduction

Fermentation of corn starch with Pichia farinosa and Klebsiella pneumoniae is a method of producing 1-3 propanediol. This substance is useful in the creation of polymers which can be woven into polyester fibers. Fermentation is a two step process with glucose being converted to glycerol and then glycerol converted into 1-3 propanediol. The fermentation produces a variety of products including Glycerol, Arabitol, Acetic Acid, 1-3 Propanediol. In order to determine the progress of the fermentation reaction it is necessary to measure the levels of 1-3 propanediol and glycerol. This is problematic because the two molecules have similar atomic structures and physical properties; therefore it is hard to separate the detection of one from the other. The method utilized in this experiment is high pressure liquid chromatography (HPLC). This experiment analyzes HPLC separations in order to determine the best methods for detecting these molecules.

Background

High Pressure Liquid Chromatography or HPLC is a chemical separation and analysis technique. The method utilizes a tightly packed column and high pressures to drive the separation. The column is primed with a mobile phase or a solution that does not absorb UV. Typically a mobile phase is selected that the various components are at least partially soluble in. In an
3

isocratic HPLC the composition of the mobile phase is constant. A gradient HPLC uses a changing mobile phase composition to drive the separation. When using a mobile phase that is more polar then the column it is called reverse phase HPLC if the column is more polar then the mobile phase it is normal phase HPLC. The idea of HPLC is using a column of material that has varying affinity for the components in solution one wants to separate. By running a sample through the column one gets a separation over the length of the column. A UV detector at the end of the column can then measure absorption due to each component. Absorption is proportional to concentration so after running a few standards the composition of the solution can be determined. A phenomenon being considered is the binding of different compounds to the column, compounds A and B. Each compound has a different affinity to binding to the column, ones with high affinity will bind to the column more readily and take longer to elute through the column. Ones with lower affinity will take less time to elute through the column. To achieve a better separation of compound A and B, one would like to lower the affinity of compound A and raise the affinity of compound B. This would allow for a better separation. The second phenomenon is diffusion of compound A and B within the column. Since there is a concentration gradient within the column, compound A and B will diffuse by Fickian diffusion. This will cause compound A and B to diffuse forward and backward at a certain flux. This phenomenon can be a problem when attempting to get better separation because the peaks will have larger bases. There are several ways to affect the binding affinity and diffusion of compound A and B by changing certain settings of the HPLC machine. A user can change the flow rate, temperature of the column, pH, use a pH gradient, sample size, column type and the mobile phase. By changing one or several of these settings, a user can increase the separation distance of the peaks and be able to determine the concentration of the sample. The amount of sample injected into the column will have an affect on the peaks base length. As more of a sample is injected into the solution will cause the base of the peak to increase due to more sample being in the stationary phase (McCalley). This effect should not affect the separation of a solution if the sample size is kept relatively small and close to the optimum sample size. Another affect of the sample size is that of overloading the column. If enough sample is ejected into the column, where all the binding sites of the column have been bound to a molecule, a plateau will occur. This will cause that run to provide no useful data and the column will have to be run till all the sample has left the column. The next affect is changing the temperature of the column will also affect the binding affinity of the compound. The binding coefficient for a compound binding to the column can be defined as:
lnK := H
o

RT

S + ln R

Where H and S are retention enthalpy and entropy, R is the gas constant and

is the phase ratio. As the temperature

increases the binding coefficient should decrease, causing a faster elution time (McCalley). This could cause a better separation if the changes in K differ more in one compound than another. This could allow a user to raise or lower the temperature of the column to create a better separation. A change in flow rate can have two different affects on the separation of a solution. The first affect occurs when the flow
4

rate is decreased. This causes the column efficiency to increase, since there is more time for the compounds to separate. The problem with this is that the peak base will increase since there is more time for diffusion to occur. Increasing the flow rate will decrease the base of the peak, but not allow as much time for the compound to bind to the column. A decrease in flow rate may cause a better separation. If none of these changes provide the necessary separation required, a change of column may be necessary. Changing the column will change the binding coefficient of each compound, which may increase the separation of the solution.

Resolution on an HPLC column is calculated with the following equation.


1 R = ( )N 2 [k ( + k )] 1 1 4
1

t N = a R W t t k= R 0 t0 = k2 k1

The value of a is dependent on the width (W) used. For W taken at the peak height a=5.54 for W taken at the tangents of the peak a=16. t 0 is the retention time of a marker chemical that has no column affinity.

Glycerol Figure 1.1

1-3 Propanediol

Equipment

Waters 600 pump Waters 600 controller Waters 2410 refractive index detector Waters Wisp 710B auto sampler Waters 2487 dual absorbance detector Shodex SH-1011 column Empower Pro IBM computer Procedure

This experiment utilized a Waters Wisp 710B auto sampler auto sampler to run sets of HPLC data. Standards solutions of glycerol and PDO between 2.5 and 12 mass percent were created. Then equal mixtures of glycerol and propanediol with two different concentrations of 2.5 and 5 mass percent were made to simulate high and low concentrations produced by the fermentation process. The mobile phase was .01 M NH 2SO 4. The mobile phase was sparged with helium at 15 mL/min. All of these samples were loaded into the auto sampler and run through the HPLC machine with flow rate varying from 0.4 to 1.00 mL/min, temperature varying from 40oC to 65 oC. The refractive index and UV absorbance were measured with 1= 254 nm and 2= 310 nm. Data was recorded and analyzed by Empower Pro on an IBM computer.

Results and Discussion

The standard solutions of glycerol and PDO all yielded single peaks on their respective HPLC runs. This single peak in each case eluted in the 28 to 35 minute region with an average peak width of 7 minutes. There was a fairly consistent spike or peak around 5 minutes in all of the samples. The standard solutions also included the 5 minute spike. Since those solutions were relatively pure except for the component inserted PDO or glycerol it seems unlikely unless the compounds were degrading that this spike was caused by another chemical in the injection solution. Neither glycerol nor PDO begin decomposing until temperatures exceed 50 o C.
6

Instead of focusing on expanding the experiment space for the glycerol/1-3 propanediol separation, this lab allowed us to become familiar with running and calibrating the Waters machine to further continue the analysis of the fermentation products in unit operations lab. Anecdotal evidence on the effects of temperature and flow rate support the literature trends of increased separation with higher temperature and lower flow rates. The best separation we managed was at our highest temperature of 65 degrees Celsius:

Figure 1.2

These two distinct peak tops are assumed to be 1-3 propanediol and glycerol, although they did not elute at the expected retention times. This retention time shift may have been caused by the increased temperature which should have decreased molecule binding affinity for the column under equivalent flow rates. Ideally spiked samples should have been run at this temperature and flowrate to determine any change in retention time from the original standards, run at lower temperature. This data was collected at 290nm and at a flow rate of 0.6ml/m. Quantitative analysis of the concentrations of the solutions was attempted, but processing challenges prevented us from obtaining statistically meaningful results. The peaks overlapped to a large degree making it impossible to get any more then a general idea of the relative concentrations of glycerol and PDO. As you can see from figure 1.2 a complete or even mostly complete separation was never achieved. The modifications of the flow rate and temperature did not cause significant changes in the binding constants for glycerol and PDO on the column. Furthermore the standard solutions of both glycerol and PDO were never run at the modified flow rates and temperatures. Since this may have modified peak areas it was impossible to get accurate concentration readings using the standards.

Future Recommendations

For future experimentation concerning this separation, we have three different suggestions. These are, changing the wavelength that the detectors used, exploration of the effects pH changes, and a change of column. The wavelengths that were used by the detectors were not the correct ones. Glycerol has a l max at 260 and 280 nm. The ones chosen in this experiment were chosen because they were used by the last lab group. Ideally ls closer to the actual l maxes of the molecules being detected would be used. Unfortunately, it was difficult to find accurate absorbance values for PDO. If different wavelengths could be used to detect only PDO and glycerol, we would be able to determine each peak in the different detectors with little to no overlap effects. This would determine the exact concentrations of each component. We recommend performing Ultra Violet Spectroscopy on standards of PDO and glycerol to determine the best wavelength to use, since literature data is hard to find. In this experiment we did not explore the affects of changing the pH of the mobile phase. In later experiments, the pH should be changed to see if it is able to see if the change of charge will have an effect on the separation. Running at high pH should cause a charge difference in the two molecules by removing the hydrogens from all of the alcohol groups (see figure 1.1 ). This would cause a charge difference that could be used to separate the molecules. A charged column would be used to perform such a separation. The exact effects on charge with respect to pH were never determined due to a lack of information on the pH of alcohols. A more in depth study of the pKas could be manually determined for the two compounds using titration. If neither one of these suggestions lead to a better separation, it can be concluded that a good separation can not be found by using this column. We would then recommend that the next lab group research to find a more suitable column. The column that was used was determined by the patent literature concerning PDO and glycerol. If a more suitable column can be found, a better separation may occur. Furthermore, the column used in this experiment is over 3 years old. At this age the column may no longer be performing at peak efficiency. Since the two compounds are so similar in nature it would only take a slight degrading of the column interior to decrease the separation to unusable levels. The separations on a new column should be tested.

Conclusion

The separation of glycerol and PDO is a difficult separation that can be improved by lowering the flow rate to generate higher pressure which generates better binding affinity to the column for the different molecules. A high temperature also improved separation. However, complete separation was never achieved but an improvement of results of the unit operations lab of 2005 was successful. We were also able to get better acquainted with the HPLC machine and all its abilities which will continue to improve the success of separations in the future.

References
History of HPLC http://www.pharm.uky.edu/asrg/hplc/history.html McCalley, David V. Effect of temperature and flow-rate on analysis of basic compounds in high-performance liquid chromatography using a reversed-phased column Journal of Chromatography A. Volume 902, Issue 2. Pg 311 321 McCalley, David V. Influence of sample mass on the performance of reversed-phase columns in the analysis of strongly basic compounds by high-performance liquid chromatography Journal of Chromatography A. Volume 793, Issue ?. Pg 31-46 Unit Operation Lab- Final Reports Corn to Polymer http://rothfus.cheme.cmu.edu/uolab/final03/ferm.html http://rothfus.cheme.cmu.edu/uolab/final04/ferm.html http://rothfus.cheme.cmu.edu/uolab/final05/ferm.html

10